As of November 21, 2026
Roadmap
What comes next for Sequences, what is in pilot now and what recent releases added. Plans can change.
Next
-
What the patent says about the matching sequence, for other offices
PlannedExpected
Publications from patent offices other than the US show their title, an English abstract where one is available, and links to the claims and the description. A machine translation is labeled as one.
-
The patent's own words in claim charts
PlannedExpected
For a US patent in a claim chart, each cell also quotes the patent's own words that bear on the feature, beside the match.
-
Fast, complete search for short DNA fragments, such as primers and probes
PlannedExpected
Exact and near-exact searches for DNA fragments of 15 to 50 bases, such as primers and probes, return every stored match on both strands, with an exact fragment found in under a second in place of about 8 s today. The search reads an index built with each snapshot.
-
Scores for each language model
PlannedExpected
Each language model that can draft claim suggestions is scored on the same published patents, chosen before the run, and the scores are published in the docs, so you can compare models before choosing one.
-
Papers that describe a claim
PlannedExpected
Sequences finds papers that describe what a claim describes, each shown with its matching sentences quoted word for word with their source.
-
Searches for proteins a claim names
PlannedExpected
When a claim names a protein instead of giving a SEQ ID NO, such as human insulin, Sequences finds its reviewed sequence and proposes a search, which runs once you accept it.
-
Result files from other sequence search systems
PlannedExpected
Import the result files that other sequence search systems write into a case. Each hit joins its patent publication and SEQ ID NO, and a row that does not join is listed with its reason.
In pilot now
-
Summaries drafted by a language model your server runs (Pilot)
ReleasedNovember 2026Released in Sequences 0.89.0
On a server whose settings name a language model, a case can ask it to draft four texts: the case summary, the summary of the results, each claimed SEQ ID NO's summary, and why each kept publication counts. Each draft names the model and prompt, every number in it is checked against the case's facts, and none counts until you accept or edit it. The hosted service runs no model.
-
Claim suggestions from a language model your server runs (Pilot)
ReleasedNovember 2026Released in Sequences 0.60.0
A case's Review step can ask the model for claim features the rules missed, with the details a search needs. None of its suggestions counts until you accept it. The case records each call. Suggestions stay off until the server's settings name a model. See Cases.
Recently released
-
Results of the US patent office's sequence search
ReleasedNovember 2026Released in Sequences 0.116.0
Import a result file reads the US patent office's sequence search results, as office actions print them: protein and DNA searches, among them those that translate DNA. Each hit joins the patent sequence its record names, and the import sits beside Sequences' own searches in References and Compare searches. See Result files from other programs.
-
Claim charts
ReleasedNovember 2026Released in Sequences 0.114.0
In a case's Report step, each claim gets a chart of its features against its top references, the ones you keep first. Each cell says Met, Not met or Not shown, with the identity, the aligned positions and the reference's SEQ ID NO, each linked to its source. You can set any cell yourself, with a note, and the charts go into the report and a CSV. See Claim charts.
-
One patent list from several searches
ReleasedNovember 2026Released in Sequences 0.111.0
Compare searches lists the patents that two to six finished searches found, each patent once, with the SEQ ID NOs and the best Query Match and identity each search found in it. Its parts show which searches found each patent, three filters narrow every search alike, and the list downloads as CSV or Excel. Patents match by office, number, kind code and publication date, even across sets.
-
Modifications on hits and patents
ReleasedNovember 2026Released in Sequences 0.111.0
Modifications on hits and patents: where a server runs the add-on, a hit whose records mark modified residues or bases shows them in the records' own words, and each patent shows what its own records mark, with filters for modified hits and for a modification you name, and export columns. From GenBank's patent sequences, EMBL-EBI's US patent proteins and the US patent office's sequence listings.
-
Stored IUPAC codes in Exact text and Pattern
ReleasedNovember 2026Released in Sequences 0.110.0
Exact text and Pattern can let a code in a stored sequence, such as R for A or G, or N for any base, match the letters it stands for, up to the most you choose in one match. The Stored codes column counts them, and the pane names each. The option is off unless you turn it on, so every other search runs as before. See Stored IUPAC codes.
-
Patent families add-on
ReleasedNovember 2026Released in Sequences 0.109.0
Patent families from Google Patents Public Data, as an add-on. See Patent families.
-
US patent status on hits
ReleasedNovember 2026Released in Sequences 0.109.0
Where a server runs the Patent status add-on, the US status column groups each US application of a hit, such as Granted or Abandoned. Its tooltip gives the US patent office's words and status date, from its Open Data Portal on the day shown. A filter keeps one group. Granted does not mean in force, and Sequences does not estimate when a term ends. See Patent status.
-
Where hits differ from the query
ReleasedNovember 2026Released in Sequences 0.109.0
Variants, beside Alignments, draws a query's first 100 hits in the query's own positions: a dot for the query's letter, the hit's own letter where it differs, and a dash for a gap. Each position counts the hits that differ and names each change, such as K45R, and Export positions (CSV) saves the table. See Variants.
-
Named identity measures
ReleasedNovember 2026Released in Sequences 0.108.0
Hit identity joins Query identity, and each measure has one name, a minimum and a sort across the page, the CSV, the workbook and the API. See Results.
-
Search from an assistant
ReleasedNovember 2026Released in Sequences 0.106.0
An assistant that supports the Model Context Protocol can list the sets, start searches and batches, follow them, and read their hits and patents a page at a time, through a program that runs on your computer with your API token. See Assistants.
60 earlier items
-
Aligned letters in the Excel workbook
ReleasedNovember 2026Released in Sequences 0.106.0
A search's workbook has a sheet with the aligned letters of the query, the match line and the hit for the first 10,000 rows of Alignments, in the same order and with the same filters. See Exports.
-
Hit by several queries
ReleasedNovember 2026Released in Sequences 0.106.0
For a query group's search, or any batch of searches, list the publications or stored sequences that at least a chosen number of the queries hit, with the names of those queries. See Query tools.
-
IUPAC codes searched as a pattern
ReleasedNovember 2026Released in Sequences 0.106.0
A DNA query with IUPAC codes can go to Pattern in one step, which reads each code as the bases it stands for. See Exact and pattern search.
-
Claimed sequences
ReleasedNovember 2026Released in Sequences 0.106.0
Which claims of a hit's US publications name its SEQ ID NO, as an add-on. See Claimed sequences.
-
The ST.26 check as its own page
ReleasedNovember 2026Released in Sequences 0.103.0
Open without sign-in.
-
Published on or after, and watches that count as new from a date
ReleasedNovember 2026Released in Sequences 0.102.0
Results, saved views, plans and the API filter by Published on or after as well as Published before, so the two dates show what came between them, such as the intervening art between a claim's priority and filing dates. A new watch can count the matches dated on or after a day as new at once. See Filters and Watches.
-
Examples of the main parts
ReleasedNovember 2026Released in Sequences 0.101.0
One page shows the main parts of Sequences working before you sign in: search, results, tools, claims and plans, model features (Pilot) and benchmarks, each with a link to a working example. Each pilot claim suggestion shows the checks that kept or dropped it, and each pilot summary draft has a link of its own. See Examples and sample results.
-
Benchmarks
ReleasedNovember 2026Released in Sequences 0.101.0
A page of benchmarks shows each recorded run of the search methods with its data, what it found and the evidence record that holds its numbers: exact and pattern counts checked against GNU grep, short DNA fragments, altered sequences, Within k changes, repeat runs, the ST.26 listing check, a search plan for every SEQ ID NO of one patent and searches an hour with more workers.
-
My work
ReleasedNovember 2026Released in Sequences 0.101.0
One page lists your searches, imported result files, cases and watches, newest first, with a filter by kind and a link back to each. On a free trial it also shows the days left, as the search page does. It lists only your own work. See My work.
-
Antisense drug names and development stages on matching hits
ReleasedNovember 2026Released in Sequences 0.100.0
A hit whose bases are an antisense or siRNA drug's, or whose chain or variable domain is an antibody drug's, names the drug and its highest development stage where ChEMBL gives one. Sequences compares the bases alone, not an oligonucleotide drug's chemical changes. See Drug names.
-
Species on each hit, and a species filter
ReleasedNovember 2026Released in Sequences 0.99.0
Each hit shows the species its record states, and results filter by species or by group, such as human, mouse or camelids, without a new search.
-
The gene behind each hit, an add-on
ReleasedNovember 2026Released in Sequences 0.98.0
In a new search, a hit on a reviewed human protein shows each gene its Swiss-Prot entry names, as 20,256 of the 20,430 human entries do. A patent protein hit identical to one shows the same genes. They appear in the hit list and the hit's pane. Like clinical context, it runs only on a server whose settings name it. See Genes.
-
Known variants: where your query differs from a human protein, each change marked when ClinVar records it
ReleasedNovember 2026Released in Sequences 0.97.0
A protein hit identical to a human Swiss-Prot entry shows the variants ClinVar records on that protein. For a protein query, each place the query differs from the protein is named, such as p.Val600Glu, and marked when ClinVar records it, with a link to the record. Like clinical context, it runs only on a server whose settings name it. See Known variants.
-
What the patent says about the matching sequence, for US publications
ReleasedNovember 2026Released in Sequences 0.96.0
A US publication's pane quotes what the patent says about the matching sequence: its abstract, the claims and up to three description paragraphs that name its SEQ ID NO, and the sequence listing's notes, each with a link to the full text. Up to three publications can be compared side by side. The quotes are stored with the snapshot, so a search calls no other site.
-
Tools to build queries from claims
ReleasedNovember 2026Released in Sequences 0.96.0
Sequence tools has a Query tools tab. Paste a phrase such as "at least 90% identical to SEQ ID NO: 1" to get proposed queries, a method and a results filter, each with its reason. Make variants from changes written as K26R, cut a sequence into windows, turn a peptide into a DNA pattern, and choose how a modified residue such as Aib is searched. See Query tools.
-
Dates for papers that have only a DOI
ReleasedNovember 2026Released in Sequences 0.96.0
A paper with a DOI but no PubMed record is dated by the day or month Crossref gives for it, where the server has Crossref's dates. The hit and the hits CSV mark that date (Crossref), and Published before uses it. See Dates.
-
Therapeutic antibody names on antibody CDR matches
ReleasedNovember 2026Released in Sequences 0.96.0
Antibody CDR search names the therapeutic antibodies a match shares sequence with: a whole heavy or light variable domain, or one chain's three CDRs in order under the search's numbering. Each name links to its page at ChEMBL, under ChEMBL's CC BY-SA 3.0 license, and the CSV lists them. See Antibody CDR search.
-
Why each publication is listed, and an X marked apart from a mismatch
ReleasedNovember 2026Released in Sequences 0.95.0
A publication's pane on Patents, and its row in a case's References step, opens with why it is listed: how its sequences match, where it falls against your date, and the earliest publication holding your query unchanged. An X in a stored protein, or an N in DNA, is marked apart from a mismatch, as in "28 of 29 identical, and 1 X", and can count as a match. See By patent.
-
Each result on its own page
ReleasedNovember 2026Released in Sequences 0.95.0
A search's result, or a group of searches, opens on a page of its own at an address you can bookmark or copy, with the search's name in the browser tab and Edit and run again to change it and search again. The search page keeps the form and your recent jobs, and older links to results open the new page. See Addresses.
-
Search settings: scoring matrix, gap costs, word size and filters
ReleasedNovember 2026Released in Sequences 0.95.0
For the method you choose, set the scoring matrix, gap costs, word size, low-complexity filter and genetic code, wherever its program allows them. Each setting shows its default and range, a value out of range is refused before the search starts, and the search record lists every setting the search ran with. Defaults stay as they are.
-
Smith-Waterman and global searches that return the same hits every time
ReleasedNovember 2026Released in Sequences 0.95.0
A Smith-Waterman or global search returns the same sequences each time it runs. A patent sequence that differs from your query only where it has an X ranks inside the 5,000 shown, and sequences made mostly of X do not crowd the list. Each hit keeps its alignment.
-
A search for every stored sequence within a set number of changes
ReleasedNovember 2026Released in Sequences 0.95.0
A new method, Within k changes, lists every stored sequence that differs from your query by at most k substitutions, or by k changes counting insertions and deletions. An IUPAC code in your query matches any letter it stands for. No rank cut or E-value applies, so the counts are complete, and each query lists up to 50,000 hits, the closest first. See Within k changes.
-
Charts, table columns, search comparison and Excel export for results
ReleasedNovember 2026Released in Sequences 0.94.0
Three charts sit above the tabs: identity against coverage, where the top hits align on your query, and hits by year against your date. You can choose a table's columns, group hits by family, applicant, office or year, see which hits two or more searches share, and export a result to Excel and each chart as SVG or PNG. The sample results show the same views. See Charts.
-
Compute for each search, and limits
ReleasedNovember 2026Released in Sequences 0.87.0
Each search records its CPU time and the data it read, the usage page totals them by account and set, and an admin can cap compute for each account and ask before a search runs past a set time. See Compute limits.
-
A case summary and one list across the case
ReleasedNovember 2026Released in Sequences 0.86.0
A case opens with a summary of its claims, features, listing and plan, and its References step with a summary of the results, each filled in by a fixed template. All references lists every reference across the case's searches, sets and imported result files once, with its rank and mark under each feature. See All references.
-
Follow-up searches from each result
ReleasedNovember 2026Released in Sequences 0.85.0
After each search and in a case's References step, fixed rules suggest up to three searches to run next from the result, such as a more sensitive method, a set not yet searched or the part of the query that no match covers. You run, add or dismiss each one, and a case's log records it. See Follow-up searches.
-
Regular expressions in pattern search
ReleasedNovember 2026Released in Sequences 0.85.0
Pattern search takes regular expressions, such as
[AG]{2,4}and(GGATCC|AAGCTT), as well as PROSITE patterns, and runs each as the PROSITE patterns it stands for, with the same limits, on both strands. See Exact and pattern search. -
Working files deleted at sign-out
ReleasedNovember 2026Released in Sequences 0.84.0
Signing out, or a sign-in that ends, deletes what that browser was still uploading or exporting, each search deletes its working files when it ends, and an admin can set how long records are kept. The audit log records each deletion with counts. See Working files.
-
Query groups and plans for claims with many sequences
ReleasedNovember 2026Released in Sequences 0.83.0
A case's draft plan groups the claimed SEQ ID NOs, says how many searches it needs and how long they take against the server's limits, runs a larger plan as several runs, and saves each locked version as a search request in Word or JSON. See Query groups and large plans.
-
Export formats your organization defines
ReleasedNovember 2026Released in Sequences 0.82.0
An admin defines export formats, with their columns, order, file type and redaction rules, and people pick one at each export. Each export records its format's name, version and SHA-256 and the file's SHA-256. See Export formats.
-
Several servers sharing one queue
ReleasedNovember 2026Released in Sequences 0.81.0
One install of the server edition runs on several servers: a coordinator keeps the queue, and workers on other servers or on the same one take its searches, so more searches run at once. A search of a snapshot keeps its place in the queue through a restart of the coordinator. See Several servers.
-
US published applications, each week
ReleasedNovember 2026Released in Sequences 0.72.0
The patent sets hold US published applications, each with its publication number, kind code, publication date, applicant, title, filing date and priority claims. Each week's applications arrive as an addition to the snapshot, so Main plus New under Collection covers the latest week. See Sets.
-
Smith-Waterman and global alignment for DNA
ReleasedNovember 2026Released in Sequences 0.70.0
Smith-Waterman and global alignment search a DNA query against the DNA sets as well as protein against protein, on both strands, with the progress line and the time left. BLAST stays the automatic choice for DNA. See Search methods.
-
Queries in PIR format
ReleasedNovember 2026Released in Sequences 0.68.0
A query in PIR (NBRF) format, whose records start with a line such as
>P1;, works wherever FASTA does: in the query box, as a dropped file and in batches for programs. The header's two letters give the query type, and a file may hold several records. See Start a search. -
Claims as filed and as amended
ReleasedNovember 2026Released in Sequences 0.67.0
A case keeps each set of claims you import, as filed, amended, published or granted, and reads claims from a Word file too. What changed compares two sets claim by claim, a locked plan names the set it follows, and the case report and the decisions file name that set. See Sets of claims.
-
An audit log for your security team
ReleasedNovember 2026Released in Sequences 0.66.0
With sign-in on, a server keeps an audit log of sign-ins, failed sign-ins, API tokens and each start with its settings. An admin exports it with every case's log as JSON lines for a SIEM, a day or a time range at a time. Each event carries the hash of the one before it, so a changed or missing event shows. See Audit export for a SIEM.
-
Summaries from fixed templates, for you to review
ReleasedNovember 2026Released in Sequences 0.65.0
A case writes short summaries from fixed templates and its own facts: of what the description says about each claimed SEQ ID NO, of each top reference and of each search. No language model takes part, so the same facts always give the same words. Every number in a summary is checked against the facts, and the report prints the ones you accept or edit. See Summaries.
-
A case's decisions as CSV and JSON files
ReleasedNovember 2026Released in Sequences 0.64.0
Download decisions (CSV) and Download decisions (JSON) in a case's Report step save every decision in the case: each claim feature, each accept, reject and edit with who made it and when, and each version of the plan. Each decision gives the number and SHA-256 of its event in the case log, so you can check the file against the log. See The decisions file.
-
Result files from BLAST+ and FASTA36 in a case
ReleasedNovember 2026Released in Sequences 0.63.0
Bring the result file of a search you ran outside Sequences into a case: BLAST+ tabular, XML and JSON output, FASTA36 tabular output, or a CSV file with a header row. Each match joins its patent publication and SEQ ID NO, and a row that does not join is listed with its reason. The file opens like one of your own searches. See Result files from other programs.
-
The search page reads what you paste or drop
ReleasedNovember 2026Released in Sequences 0.61.0
Query type reads DNA / RNA or Protein from what you paste, and Search in follows it. A dropped FASTA file fills the query box, a sequence listing opens its SEQ ID NO picker, and a publication's XML with claims starts a case. Each result has an address that opens it again. See Query type.
-
References under each claim feature, ranked by fixed rules
ReleasedNovember 2026Released in Sequences 0.59.0
A case's References step, between Run and Report, lists every reference its searches found under the claim feature that found it. References rank by fixed rules: whether the best match meets the claim's threshold, how much of the claimed region it covers, its date against the prior art date, then Query Match and E-value. See The References step.
-
Search several sets at once
ReleasedNovember 2026Released in Sequences 0.57.0
Check several sets under Search in and they run as one combined search, followed in one status bar. The result opens on one list of references, with a chip for each set that found each one. See Search several sets at once.
-
Listings inside a publication's full-text XML
ReleasedNovember 2026Released in Sequences 0.56.0
Import in Sequence tools reads the sequence listing inside a patent publication's full-text XML, as in publications of 2001 to 2004. Programs can also find their jobs by name, set or state. See Listings and the ST.26 check.
-
Sample results before sign-in
ReleasedNovember 2026Released in Sequences 0.55.0
Real searches of public sequences, saved and shown read-only, that anyone can read before signing in. The sign-in page links to them when a server has one. See A sample result.
-
Suggested searches
ReleasedNovember 2026Released in Sequences 0.54.0
The search page offers real public sequences, such as GLP-1 and human insulin's gene, that run with one click. Their results show two or three things to try next. See Suggested searches.
-
Patent families on Patents
ReleasedNovember 2026Released in Sequences 0.53.0
With Families under Group, each row on Patents is one patent family, kept or excluded as one. With Published before set, the later publications of the same sequences fold into one line under the list. See By patent.
-
One search across several sets, for programs
ReleasedNovember 2026Released in Sequences 0.52.0
POST /api/groupsruns one search across 2 to 8 sets and gives one list of what they found, where a patent or database entry that several sets found is one row. See Programs and API tokens. -
Results as one-line rows beside a detail pane
ReleasedNovember 2026Released in Sequences 0.51.0
On Patents and Alignments, each result is one line. Select a row to read all of it in its pane, mark it and move to the next without leaving the list, with the mouse or the keys. See Results.
-
A case's prior art date, and plans that search every set
ReleasedNovember 2026Released in Sequences 0.50.0
A case keeps its application number and prior art date, and new plan searches look for what was published before that date. A drafted plan searches every set on the server that holds the finding's type, and a locked plan of any size runs to the end by itself. See Cases.
-
Earliest publications first, and filing dates in the date filter
ReleasedNovember 2026Released in Sequences 0.49.0
Publications that tie on a sort list the earliest first. Published before also keeps a sequence when one of its US or WO publications was filed, or claims priority, before your date. Programs can list a search's patents by family. See Filters and sorting.
-
Funding trail, an add-on
ReleasedNovember 2026Released in Sequences 0.46.0
The papers on a hit list the NIH awards their grants match, each linked to NIH RePORTER, and their other grants as PubMed lists them. The page draws no conclusion from them. Like clinical context, it runs only on a server whose settings name it. Since 0.48.0 it can also show what a hit's US published applications state about federally sponsored research.
-
Clinical context, an add-on
ReleasedNovember 2026Released in Sequences 0.45.0
A protein hit that is a drug target shows the drugs that act on it, the diseases they are studied for, the highest clinical stage and the clinical trials, each linked to its public source. An add-on runs only on a server whose settings name it, and a server that names none holds none of its data.
-
Install kit for AWS
ReleasedNovember 2026Released in Sequences 0.43.0
A CloudFormation template and a script that set up the server edition in your own AWS account: the server, its disks, its network rules and its role, and with sign-in an HTTPS load balancer. The server copies each snapshot from your S3 bucket and checks it against its SHA-256 list before the service starts. See In your AWS account.
-
Applicant watches and portfolio export
ReleasedNovember 2026Released in Sequences 0.42.0
A watch can follow an applicant as well as a search, and lists the applicant's new publications after each weekly addition. You can export one applicant's publications with their sequences, and the case report has a section by applicant. See Watches.
-
Draft sentences for each kept reference
ReleasedNovember 2026Released in Sequences 0.41.0
For each publication a case keeps, the case drafts up to four sentences from stored facts: the alignment, the claim's identity threshold, the publication date against the filing date, and the genes and diseases the paper names. You accept, edit or reject each sentence, and the report prints the ones you accept or edit. The same facts always give the same sentences. See Cases.
-
Paper dates and citations from PubMed
ReleasedNovember 2026Released in Sequences 0.40.0
Each paper on a hit reads as its PubMed citation, with its PubMed ID and DOI, and is dated by the day it first came out, online or in print. Published before and a hit's earliest date use that day instead of the year alone. See Papers.
-
Retraction marks
ReleasedNovember 2026Released in Sequences 0.40.0
A paper that was retracted, drew an expression of concern or was reinstated shows a mark with the notice's day and its source: PubMed, the Retraction Watch Database, or both. The paper keeps its place and its date.
-
Results by applicant
ReleasedNovember 2026Released in Sequences 0.39.0
The Applicants tab groups a search's patents by applicant, with each one's publications, earliest date and best identity. Spellings that differ only in case, accents, punctuation or a legal ending such as Inc or GmbH share a row. You can join names that still differ, and a row can link to the applicant's record in the Research Organization Registry. See By applicant.
-
API tokens and batches for programs
ReleasedNovember 2026Released in Sequences 0.38.0
Personal API tokens for programs, batches of up to 100 searches, and an OpenAPI 3.1 description of the routes a program uses. See Programs and API tokens.